From: remitamine Date: Fri, 11 Sep 2015 03:44:17 +0000 (+0100) Subject: [utlis] add extract_attributes for extracting html tags attributes X-Git-Url: http://git.bitcoin.ninja/index.cgi?p=youtube-dl;a=commitdiff_plain;h=689fb748ee1ba8e61f99d21a3bcb1bc83b708649 [utlis] add extract_attributes for extracting html tags attributes --- diff --git a/youtube_dl/utils.py b/youtube_dl/utils.py index 206dd56bc..bcebf9cc5 100644 --- a/youtube_dl/utils.py +++ b/youtube_dl/utils.py @@ -248,6 +248,14 @@ def get_element_by_attribute(attribute, value, html): return unescapeHTML(res) +def extract_attributes(attributes_str, attributes_regex=r'(?s)\s*([^\s=]+)\s*=\s*["\']([^"\']+)["\']'): + attributes = re.findall(attributes_regex, attributes_str) + attributes_dict = {} + if attributes: + attributes_dict = {attribute_name: attribute_value for (attribute_name, attribute_value) in attributes} + return attributes_dict + + def clean_html(html): """Clean an HTML snippet into a readable string"""